A team of Australian researchers, including a Monash University expert, has developed new AI-driven models to help policymakers 'pick winners' and maximise scarce resources when investing in technologies to protect threatened ecosystems such as the Great Barrier Reef.
The study, published in Proceedings of the National Academy of Sciences, applies adaptive management techniques and mathematical models to address one of the toughest policy challenges: determining when to persist with technology development despite risks, uncertainty and high costs.
The AI mathematical models identified the optimal stopping point when continuing development of an ecological technology was no longer worthwhile, and investment should be redirected to other conservation projects, accounting for the uncertainties and complexities of protecting threatened ecosystems.
Co-author of the research, Monash Environmental Informatics Hub Director Professor Iadine Chadès, said the study built on two decades of research in artificial intelligence for conservation decision-making.
"My work has focused on developing AI-based frameworks particularly using partially observable Markov decision processes to guide adaptive management under deep uncertainty," Professor Chadès said.
"This project extends those methods to a new frontier; planning technology development itself.
"By applying AI decision frameworks to research and development, we can help governments and conservation agencies make smarter, more transparent investment choices.
"It's about turning mathematical insights into practical strategies that improve the odds of saving ecosystems."
The research was led by Queensland University of Technology (QUT) PhD candidate Luz Pascal, who said the model would help decision-makers determine how long to keep investing in a particular technology before it was time to move on.
"Research and development is inherently risky," Ms Pascal said.
"We might invest millions into a technology that ultimately never works, but the potential benefits could be enormous.
"We might be able to protect an ecosystem that was previously thought to be damaged beyond repair.
"To navigate these uncertainties, we've created a way to calculate the 'optimal stopping point' - the moment when it's no longer worth continuing development of a particular technology.
"This finding is very important when we have limited resources to manage ecosystems: there is a point where we need to divert our investments to other conservation actions.
"And we find that sometimes, it is best to not invest in the new technology at all.
"Our study delivers clear and transparent general rules for investing in new technologies, building on an analytical approximation."
The team found that the model showed a clear "time limit" on how long decision-makers should invest in developing technologies before cutting losses.
QUT Professor Kate Helmstedt, who conceived the project, said the research was the first to integrate both the development and deployment phases of technology into a single optimisation framework for biodiversity conservation.
"This optimal stopping point depends on the ecosystem's characteristics, the impact of the technology, and how confident we are that the technology will succeed," Professor Helmstedt said.
"When protecting threatened ecosystems these elements are all complicated and uncertain.
"This study's novel approach to planning research and development for environmental technologies was to balance the uncertainties and rewards of innovation in ecological contexts."
Using the Great Barrier Reef as a case study, the team found that the optimal development time for new technologies could range from 0 to 45 years, depending on the confidence in their effectiveness.
"This work is a game-changer for how we think about investing in innovation for conservation," Professor Helmstedt said.
"It's not just about developing the next big idea; it's about knowing when to pivot, when to persist, and how to maximise impact with limited resources.
"Our framework is flexible and can be adapted to other domains like public health, energy, and climate resilience.
"It's a powerful tool for strategic decision-making in uncertain environments."
The research team comprised Ms Pascal, Dr Matthew Adams, Professor Kate Helmstedt, all from QUT; and Professor Iadine Chadès from Monash University.
The research was supported by the Australian Research Council and contributes to the Monash-led Securing Antarctica's Environmental Future research centre.
Read the paper online: https://doi.org/10.1073/pnas.2422002122
